Average is Over, computer skills edition

Summary: Across 33 rich countries, only 5% of the population has high computer-related abilities, and only a third of people can complete medium-complexity tasks.

That is from Jacob Nielsen, via Roman Hardgrave.


There's also a steep fall off in ability to learn new computer programming languages and the like after, maybe, age 40.

If there is a steep decline in learning new programming languages after around 40, then it should be easy to find a study to back this claim.

There is somewhat of a decline after 40 in language learning on the whole but not "steep."

I suspect that's much more true for people who start programming later in life, or who are well-established enough in a particular technology that they have little incentive to learn another. Seems like the older I get, the easier it is to learn new programming languages and techniques (except for html tags, apparently). That might just be because modern programming languages are so much easier to learn and use due to the proliferation of tools and training, but overall theoretical knowledge of programming is always useful in learning new material, and should never decline much absent health problems.

As I approach 40, I find myself having less *desire* to learn languages and frameworks. I know rationally that I should, so I work at it when I can, but the excitement isn't there.

I need to break out and learn some things likes Haskell and Prolog so that I can tackles those kinds of programming before I ossify completely.

Heartily agreed. I'm north of 50, and can still learn a basic c-based OO language easily, but start talking about paradigms that I never learned in university (functional programming, closures, etc.), and it takes a *lot* of work, which means the payoff has to be *really* high.

And sadly, sheer novelty isn't quite enough payoff anymore.

I may never learn Swift, which I would have certainly picked up 20 years ago, even if I never touched an iOS device.

Start with Gulliver's Travels, see how you feel, then if you are hungry for more maybe skim A Modest Proposal.

Learning new languages kinda sucks, and takes weeks and weeks of frustratingly making seemingly no progress on mundane shit. I'm switching from R to Python now, and have spent 14 days doing what I could have done in 1 day in R. I imagine if you're 50, if you've managed your career correctly, the marginal return towards investment in weeks/months of frustrating learning vs. opportunity cost, is not the same as for me, a guy in my 20s.

Well played, sir.

This is optimal explore/exploit behavior according to what I recently read in this:


After your first 10 or so languages, you realize new ones are just old languages repackaged. Sure, if you want I'll "learn" this language for this new job, which basically means applying one of my generic language filters and then looking up "how does this library make me do substrings" as it comes up.

Ah, substrings. I probably have notes on parsing strings in 20 different languages.

Anything you can do in Ruby or Python I can do in 8051 assembly language.

8051? So, the legends are true, then?

If one has never encountered functional languages before, and sees F# for the first time, this can be quite a shock. Definitely not a repackaged Java.

What is the suggested cause behind the "steep fall in the ability to learn a new computer programming language"? a) our brains decay, b) we simply specialize into career field.

I'd go for option b. If you specialize in general purpose code like C++ or Python, you won't quit your job at 40 to start making web pages with HTML or Java. Maybe as hobby but not as income generator. If you work in science or engineering, you stay with Fortran, Matlab, C++ or R. It's an interesting question. Programming languages life is longer than most people careers. Fortran and SQL are damned old. C++ and Matlab are 30+ years old. Java, R, Python, HTML are 20+ years old.

What Steve comments is something my father experienced in the 80s in the pre-Microsoft dominance era. They got IBMs with PC-DOS along BASIC. Then they guy in in charge of the computer department got impressed by SUN and they bought workstations with SunOS (Unix). Then Lotus 1-2-3 was the new shining object and they went back to IBM. Once Microsoft Office became dominant in early 90s, the industry & business people just had to learn Excel and PP. Most of engineering applications exist in Windows. There's some specialized or open source software for Linux but people working in Linux have no learning problems.

In conclusion, 20 or 30 years ago the average worker had to learn "several programming languages along his career" (in reality: operating systems) because the software market was precisely in the experimentation and development phase. In the early 90s, it settled and learning a new language (or OS) is a task taken by the people that have no problem learning new stuff: coders.

I am a computer scientist, and this is incorrect. What is true is there becomes competing incentives - advancing and promotions tend to rely on deeper specialization, and so rather than learn new languages and techniques, there's greater incentive to reach increasing levels of mastery within an existing language or toolset.

However, due to the radically disruptive nature of technology, every five to seven years something new comes along that renders previous skills and mastery obsolete. So the previous Java / WebSphere master now needs to start over learning Ruby on Rails, which college graduates have been using for years in school and side projects. If you're an employer presented with a Lvl 3 25 year old vs a lvl 1 40 year old, you choose the lvl 3 - higher skill, more potential.

It has nothing to do with ability to learn and age. Just timing.

What is true is there becomes competing incentives – advancing and promotions tend to rely on deeper specialization,

Yes. This is the "paradox of specialization".

Specialization benefits the worker, the company, and the economy in general. But when a technological area ceases to grow (for example when the technology becomes mature and ubiquitous), the worker is left unemployable and gets sneered at by the likes of TC who contend that the economy is essentially in a situation of full employment.


....the referenced international "computer skills" study has nothing to do with advanced skills like programming --- that study addresses very basic computer skills for average people.

"The main point...is that ... the top category of computer skills {Level 3 in U.S.} only 5% of the population has these high computer skills."

These Level 3 skills (as defined by these researchers) are relatively simple compared to programming skills--- they involve end user tasks like extended web searches with browser and multi-step email communications. The skill levels are more a measure of logical thinking abilities rather then computer-unique skills.

The data, http://www.knosof.co.uk/ESEUR/figures/index.html#Robles_14, shows that average of first contributing to Open Source is growing. That could be dues to more older programmers not having the opportunity to code at work (older people tend to be paid more, its cheaper to get younger people to do it).

I grew up on Fortran and PL/1. Graduated to a high-level systems programming language, and created a 4GL for financial applications, At 50, I moved into Visual Basic and SQL. At 58, I jumped into C#. I learned it over a weekend for a job interview. At 64, I added HTML, CSS, Javascript and jQuery to my repertoire. I led a team that converted 1.3 million lines of Visual Basic front-end code to HTML and Javascript and created a better application. My motto was to stay away from the bleeding edge, but go with the leading edge.

I don't think you lose the ability to learn new technology if you deliberately learn how to learn new languages. OTOH, I've know programmers who learned COBOL on mainframes and never escaped.

Average is *extra* over in Japan.

Not really.

Japan's computer market has been different than the U.S. and I assume Europe since the 80s and 90s but is I think with smartphones and ipads is converging.

In the 80s and 90s, the fax machine was far more common than the U.S. and many homes had stand alone word processors and video game counsels. With the older generations, many didn't see a use for a computer when they had a good word processor and their trusty fax machine. Computers were much more expensive in Japan than the U.S. and that seems to have lasted until around 2000/2001 when I saw PC/laptop prices drop noticeably and broadband took off, although the internet was very expensive for many in 1997 to 2003 as local phone hook ups were 10 yen (cents) a minute.

Then the iphone became popular in 2008 and much more so in 2009 so that was another device followed by the ipad in 2010 where the laptop could be sidestepped. The ones that have computers probably do now how to use them better than Americans and others, although the gap isn't that large.

I am a senior professional programmer at Fortune-100 software company and I'm not sure how well I would fare on their example tasks, which mostly involve email interfaces.

Time was, an email program would present a query interface that allowed you to explicitly construct clearly defined search criteria on clearly defined fields, even connected by AND and OR operators. If the program was any good, you could export your results as text for further processing by other programs. Nowadays, in the name of user-friendliness, there is typically an empty search bar. There is no documentation saying how what you type is transformed into a query, because that would be user-unfriendly, and anyway what query is produced is likely to change every week or so as the server-side logic is updated, unbeknownst to you. The query might depend on what you have browsed for over the last few days, or what an advertiser paid your email provider last quarter. Since the whole thing is probably backed by a distributed No-SQL system, there is no guarantee that you will get back everything that matches anyway. And forget ever being able to export anything, since that would undermine the lock-in on which your provider has based his business model.

+1 there has been an unfortunate trend in removing features for power users in favor of ease of use and "clean" design (which means hiding everything).

It's frustrating, because in 2016 nearly everyone should be "power users", given the amount of time most people spend interacting with computers. But they aren't given the tools to work more efficiently and effectively.

p.s. I'm under 40.

This is not only true for software, but pretty much any mechanical or electric device. Open the hood of a 1970s car and you will understand what is going on. I remember a physics teacher in high school that would take the students to the parking lot, and teach some physics concepts from the engine of his 1980 Ford Something. Open the hood of a car today and you see plastic and electrical connectors. The general pattern is clear, all these products are becoming increasingly complex, as well as more user friendly. Power users are less numerous than we think.

If you don't understand how a modern car works, don't blame car companies. A low skill blue collar can connect an OBD2 reader into the port and know what happens with the car. Any car enthusiast can learn to play with the computer car. You may void the warranty if you connect your laptop to the OBD2 port and you may break some things while learning......but that's what car enthusiasts did in the 50s. They destroyed engines and void warranties too ;)

" low skill blue collar" Is there any other kind? Only white collars, orthographic conceptualists, are high skill. Object manipulators are low skill because they don't deal in abstractions and the objects they manipulate actually must function.

That's not his point. His point is that with older cars you could see the discrete parts and explain how they related to each other and the functioning of the car. Now when you open the hood of a car you can't really see much. Engine bays are more compact, crowded and harder to work on.

When I pop the hood on newer cars, I now find the engines hidden under plastic covers. It's a bit disconcerting, and also sad that car makers even don't want to show off anything under the hood.

Well, when I open the hood I can still see where all the parts are: brake pump, radiator, steering pump, oil filter, ECU (the computer), AC compressor, transmission, turbo, alternator, etc. If you follow the cables you can see where the sensors are. In compact cars due to reduced space it's easier to work from below the car. That's why at mechanic shops they have car lifts.

Before moving to Europe I had a 4x4 Ram diesel and I had fun working on my truck. Yes, lot's of cables, sensors and plastic covers. I think the guys that say "you can't understand how a modern car works" have never actually worked in a modern car. It's about cleaning, changing parts and put everything together again. Explaining how the parts relate with each other is for men that like to explain but do nothing.

Computer use doesn't necessarily lead to super-use anymore than handling paper will turn me into a novelist. Searching through Amazon, pinning stuff on pinterest, scrolling through Facebook...these things do not make a super-user.
I can't even create a proper SQL query, but there are plenty of GUIs that handle it for me.

In the newest iteration of MS Outlook, it appears that there is no longer an option to search for exact phrases. If you use a search string like "Q1 sales", you'll get all emails with the words "Q1" and "sales" in them, even if they're not written together. I don't understand who would think this was a good idea.

As a business application developer, we do some work to dumb down interfaces for lower level workers who lack these skills, but mainly we make them less painful and more efficient -- honestly at the end of the day the users themselves are the source of most of the requests for logical complexity, which can be extensive, and our value to them is to deliver an application that supports the complexity of their business processes.

Th numbers for the 4 skill levels don’t sum to 100% because a large proportion of the respondents never attempted the tasks, being unable to use computers. In total, across the OECD countries, 26% of adults were unable to use a computer.

Wait, what? They couldn't log in? Didn't know the password? Didn't know how to use a keyboard? Didn't want to learn how to use Windows? Didn't care? You have to wonder to what extent the authors may have confused "unable" with "unwilling" or at least "unenthusiastic."

My kingdom for an edit button.


But pressing it would presumably be a medium-complexity task and therefore unsuitable for end-users.

"... honestly at the end of the day the users themselves are the source of most of the requests for logical complexity,"

Yes. I spend a good part of the design process pushing back against requests for more complexity. Quite often it's driven by the expert operator. However, if you let the expert operator design the GUI, the average operator will be overwhelmed.

So, generally, I try to push for "Maintenance" or some kind of high level screen that can be used for advance functionality, but doesn't clutter up the Main screens with non-primary functionality. Of course, operators hate to have extra clicks, so that's a battle too.

"Operator: What you mean it's 3 clicks instead of 2 clicks! Me: How often do you perform this function during a shift? Operator: Oh, at least 2 or 3 times. Some days, as many as, half a dozen! Me: Hmmm, I see your point. Ok, how about we try it out on the Maintenance screen first, and then if it because an issue, I'll move it to the Main screen during the Field Support phase? Operator: Ok, sure."

As a business application developer, we do some work to dumb down interfaces for lower level workers who lack these skills, but mainly we make them less painful and more efficient -- honestly at the end of the day the users themselves are the source of most of the requests for logical complexity, which can be extensive, and our value to them is to deliver an application that supports the complexity of their business processes.

Th numbers for the 4 skill levels don’t sum to 100% because a large proportion of the respondents never attempted the tasks, being unable to use computers. In total, across the OECD countries, 26% of adults were unable to use a computer.

Wait, what? They couldn't log in? Didn't know the password? Didn't know how to use a keyboard? Didn't want to learn how to use Windows? Didn't care? You have to wonder to what extent the authors may have confused "unable" with "unwilling" or at least "unenthusiastic."

Yeah, and likewise, how many of the people who actually did start the tasks failed to complete them for similar reasons?

" 26% of adults were unable to use a computer. ... You have to wonder to what extent the authors may have confused “unable” with “unwilling” or at least “unenthusiastic.”"

That's pretty much the 90 IQ and less range. You are going to see a significantly reduce capacity to deal with complex tasks. It's unreasonable to expect 100% of the population to be able to use a high complexity device.

Some computer tasks are high complexity, but computers also use a lot of low-complexity applications, and IQ is knowledge divided by age. Hard to believe a quarter of the adult population is incapable of reaching the level of computer savvy typical of 10-year-olds.

I was a decent programmer at 10, and it was all downhill from there.

"Hard to believe a quarter of the adult population is incapable of reaching the level of computer savvy typical of 10-year-olds."

20 to 25% of the US population fails to graduate from High School.

Again, though, how badly did they want to?

Also, HS Grad or equivalent approaches 90%, according to this source.


IQ 90 is roughly the average IQ of the world's population.

Are these users also unable to use a contemporary TV? Or can they use a TV yet remain unable to access many of the TV's features?

A TV is no longer a box with a channel-selector knob on one side and a volume/off knob on the other; it's become a type of embedded computer system.

As have many common appliances. Even without IoT/connected home mania, are we approaching a point at which many people are only able to make limited use of their appliances? And if so, is this inherent in the nature of these things or is it more a failure in design?

"only 5% of the population has high computer-related abilities": that includes me - to agree, you'd only have to see the stuff I produced when I was young.

"and only a third of people can complete medium-complexity tasks": I may not be in that third, because I get bored and frustrated by most of the superficial rubbish in "tech", involving as it does wrestling with the stupidities of the programmers. The trouble with IT is that it's full of modestly clever people who think they are very clever. See Nielsen's first para for confirmation.

These statistics means nothing until we understand what "high computer related abilities" and "medium-complexity tasks" are defined as.

The article gives examples. It doesn't have anything to do with computer programming, and seems to be mostly business-type activities like using Outlook.

I wonder where using Google or buying something off Amazon ranks in their difficulty levels. It can take some persistence and reasoning to get what you want. Yet most users seem to manage.

I like your metric.

But, your metric is consistent with the results of the OECD research. 65% of the US population managed to buy something online on 2015. This is consistent with the percentage of level 3 + level 2 + level 1 tech skills = 65% for the US. I'm impressed.


Is this 65% of adults or the entire US population? I can't access that site, but this one says 70% of US adults shop online at least monthly: http://www.mintel.com/press-centre/technology-press-centre/nearly-70-of-americans-shop-online-regularly-with-close-to-50-taking-advantage-of-free-shipping

That suggests the number of people with at least level 1 computer skills is quite a bit higher than 65%.

How does Facebook even make money? I wonder how many Americans have strong auto repair skills? How many could change an oil filter?

"The meeting room task described above requires level-3 skills. Another example of level-3 task is “You want to know what percentage of the emails sent by John Smith last month were about sustainability.”"

I'm pretty strong with computers but I'd have trouble with this. As far as I can tell Outlook does not have great counting and statistical features. I could easily search out all John Smith emails sent to me then I can search out emails he sent about 'sustainability' but to get the percentage I'd have to count them manually.

It would be cool if Outlook had a feature to accumulate data, average emails sent per day, per hour, received, etc.

"“You want to know what percentage of the emails sent by John Smith last month were about sustainability.””"

"I’m pretty strong with computers but I’d have trouble with this."

If you have issues with that skill then you are not pretty strong with computers.

MSOutlook, Search for all emails From:"John Smith"
MSOutlook Search for all emails From:"John Smith" AND about:"sustainability"

Divide B by A

That requires learning how Outlook works today, a skill which will be useless in 3 years because at least one of these will happen: Outlook will change; my email client will change; my employer will change to one that uses Google Mail For Business.

Well, there's a search bar prominently positioned in the interface, which has been there for several years and is probably not going anywhere because many people use it a dozen times a day. There's also a help feature that explains how to use it.

Depending on the version, I believe Outlook will tell you how many results you got.

I can do both of those searches easily but does Outlook count the emails? If I'm dealing with multiple screenfulls of emails do I have to count them myself?

Does finding the word sustainability in the email indicate it was about sustainability? If outlook has a feature where a sender can tag an email by topic I might understand, but unsure if this is done. It's been several years since I've used Outlook.

.... or does an email not having the sustainability mean it was not about sustainability?

I would rather pay someone else to change my oil filter, but if I had to change an oil filter, I would first Google up a video on it, and I'm confident after a significant amount of cursing I would eventually accomplish the task.

"I would rather pay someone else to change my oil filter, but if I had to"

This does, I think, belong to a class of skills where one can trade time-to-completion for skill.

If this is something you do every day you'll get so you can do it quickly and with good quality. But if you don't, you can still do it slowly by following step-by-step instructions, trying to avoid doing anything you can't undo, and paying careful attention to detail.

The problem with extending this approach to computer tasks is, there are so many of them and in so many variations that this trading time for skill won't work nearly as well as it might for manual skills. And physical things tend to remain consistent in how they respond to input, unlike software where operating a control will often produce very different outputs, depending on context.

Although I don't doubt that some people do approach common computer tasks using step-by-step instructions. At least until some smarty changes the user interface. Again.

Well, I worked in an engine shop in college, if mainly as the parts/computer guy. I've never really considered changing my own oil, but I suppose I don't want to get dirty, or keep a lot of tools or oil around, plus the hourly rate I pay someone else to do it is somewhat less than I make after taxes so I come out ahead by working instead of changing my oil.

Actually, now that I think about it, my IT challenges are also mostly met via the combination of Google and cursing.

But once you've figured out how to do something once, and it's in your notes, it's all return on investment. Soon you've built whole architectures that deliver enormous value to the business, that no one else understands or can support, and in record time and budget.

And a couple weeks after implementation, when it runs well enough to not need much support, they send you off to learn something else.

My friend Matt has what I call "useful skills": he can build a house, repair a clogged drain, navigate a large sailboat, fix a motor, to name a few. Me, on the other hand, I have what I call "useless skills": I can draft documents for a complex transaction, give advice on the legality of a proposed course of action, negotiate a business transaction, to name a few. Where do the skills of a computer programmer or hedge fund manager fit, useful or useless?

Those aren't useless skills. Drafting a good legal document takes skill and expertise. Source: accountant who needs to collect unpaid charges as a result of bad legal documents and miscommunication.

Given that it is a joke, it depends on whether you want to extend the joke to those classes of people or not.

"Where do the skills of a computer programmer or hedge fund manager fit, useful or useless?"

Well, during the revolution, the hedge fund managers are probably third against the wall. After the politicians and the Lawyers, of course.

I always thought the lawyers would lead the revolution.

Have you noticed who's in charge?

Could just look at the hourly wage equivalent, no? Just make sure you compare likes, simple bankruptcy vs calling a plumber for a clogged drain (although, flat chareg wise, I guess that's about 500 vs a grand)

Are we talking about EMP apocalypse takes out all energy and communications systems sort of scenario?

Or maybe "useful" can be defined a little bit more in relation to non-apocalyptic scenarios ... (if you can apply those skills to reducing the probability of apocalypse, that's pretty awesome too).

This feels off to me. For example, just think about the percentage of people in the US who have managed to figure out how to use Facebook.

People can use Facebook at a rudimentary level but many don't know how to use features. Google found that 90% of users don't know how to use Ctrl-F and 50% of high school teachers that are supposed to teach students don't know how to use Ctrl-F, either. I've used it on average 10 times each day for the last 30 years because it's an essential function not only of web browsers but many other apps such as word processors and spreadsheets.


Amazing that the words "phone" and "tablet" do not appear above.

We are in an interesting time. Kids are increasingly given phones and tablets at very young ages, and this gives them the "bravery" to pound buttons on a new app until it does what they want. This is what is specifically lacking in the over-somethings. This seems to be what the article thinks are "computer skills."

On the other hand, fewer and fewer kids use general purpose computers. They may go from phones and tablets to XBoxes and Chromebooks. "Computer programming" like "robotics" becomes a middle school special segment. "Everyone codes" was tried and I think died.

We will have skilled app users because that is now native. And then we decide if we care that not everyone codes.

"One of usability’s most hard-earned lessons is that you are not the user. This is why it’s a disaster to guess at the users’ needs. Since designers are so different from the majority of the target audience, it’s not just irrelevant what you like or what you think is easy to use — it’s often misleading to rely on such personal preferences." (from the link)

And that seems to produce these types of results.

Not really sure how that relates to the average being over -- from a UI design point I'm pretty sure we never got to a good average to begin with. Costs have largely been sifted to users because the designer or programmer didn't want to do the better quality work but wanted to pump something out to market blissfully ignorant of what was needed/wanted but the users. However, many seem to have learned how to make that work for them in a lot of settings (both work and home).

"The goal of the problem may have to be defined by the respondent, and the criteria to be met may or may not be explicit" while this appears to be a part of the skills for the top skill level I fail to see how it's actually a computer skill -- that's analytical and organizational and reading skills not computer skills. Moreover, it's dependent on the way the information was provided so not independent of the test itself.

"Not really sure how that relates to the average being over —"

I forgot to add that. I think you could find something similar going back to the 1990s and likely the mid 1980s.

That means software needs to get better.

Strange. I always thought the IT sector was oversaturated. So many guys graduating with degrees from low tier Universities, or teaching themselves to build websites, etc.

Very few people know the very useful skill of heart transplant. I remeber when specialization of labor was at the heart of modern economy. But maybe it' just me.

Skills are required in any part, if we have skills then only we will be able to do anything worthy, if we lack the skills then nothing will be good. I mainly focus on trading career where I get it done rather nicely through OctaFX broker which is absolutely world class been regulated plus a true ECN as well, so that’s why trading with them is such a pleasure and helps in every way while keeps everyone entirely comfortable with things.

Comments for this post are closed